Evaluation of empirical type I error and empirical power using the exact method for the cholangiocarcinoma study with months and effect size equal to improvement in median time by a factor of 1.5 - using 10,000 simulations (nominal type I error 5%, target power 80%).